Re: sqlite3

[Qiantan Hong]

> On Dec 12, 2021, at 6:39 AM, Teemu Likonen <tlikonen@iki.fi> wrote:
> 
> * 2021-12-12 14:10:28+0000, Qiantan Hong wrote:
> 
>>> Which is exactly what is being planned. No one has any intention to
>>> go over all the data Emacs stores between sessions and convert all of
>>> them to sqlite3. We only plan to use that where a data base is large,
>>> and thus using plain text there is unwieldy.
>> 
>> I’ve implemented exactly this, and your multisession.el, with a little
>> modification, would fit into this plan and functions as a key-value
>> store backend.
> 
> Or maybe another approach: Use multisession.el as the general front end
> interface which could dispatch to different key-value storage
> implementations, defined in other files like multisession-sqlite.el,
> multisession-ihatesql.el, multisession-postgresql.el etc.
I think it’s much easier to use it as a backend, as it just require some
slight modification to do so. It is also closer to a key value store
than persistent variable (I know, it is close to persistent variable,
but persistent variable is also closed to key value store and 
multisession.el is closer to key value store between the two).
Just compare
(multisession-value x) with (store-get x multisession-store).
I also think changing the interface to the latter is useful,
because the user can then use the database of their choice.
It can default to a global one.